Application of the atomic force microscope to integrated circuit reliability and failure analysis

Leakage current study and relevant defect localization in integrated circuit failure analysis

This presentation is an introduction to machine learning techniques and their application in semiconductor failure analysis. The presentation compares and contrasts supervised, unsupervised, and reinforcement learning methods, particularly for neural networks, and lays out the steps of a typical machine learning workflow, including the assessment of data quality. It also presents case studies in which machine learning is used to detect and classify circuit board defects and analyze scanning acoustic microscopy (SAM) data for blind source separation.

A vacuum chamber for a commercial atomic force microscope (AFM) is designed and fabricated. Only minimal modifications were made to an existing microscope in an effort to work in a vacuum environment, while most of the available AFM functionalities were kept intact. The optical alignment needed for proper AFM operations including a SLD (superluminescent diode) and a photodiode can be made externally without breaking the vacuum. A vacuum level of 5×10 -3 torr was achieved with a mechanical pump. An enhancement of the quality factor was observed along with a shift in the resonance frequency of a non-contact-mode cantilever in a vacuum. Topographical data of a calibration sample were also obtained in air and in a low vacuum using the non-contact mode and the results were compared.

This study presents the design and implementation of a prototype device that interfaces with commercial atomic force microscopy (AFM) instruments to accurately measure ex vivo, regional, passive elasticity measurements of the murine myocardium in three-dimensions (3D). The constructed prototype has a translation stage base allows fine adjustment of planar location at the micrometer level of accuracy. Four degrees of motional freedom are allowed in total, permitting fine, independent rotations of the murine heart (oriented at right-angles with respect to its normal mode of operation) at 2–5° increments, along its two axes of symmetry. Successful mounting of the constructed prototype was realized on the AFM’s base platform, with Zpositioning maintained through the instrument’s electronics. Measurements of normal epicardial tissue elasticity in one excised male C57BL/6 heart led to a mean value (over 15 epicardial locations) of 51.8 ± 5.4 kPa. A flexible, versatile, and low-cost prototype assembly was designed, implemented, and tested to conduct epicardial elasticity measurements in a semi-automatedmanner, using AFM. The epicardial elasticity values are within expected, published ranges for the normal myocardium of 10–55 kPa. Efforts are ongoing to allow 3D regional elasticity measurements at finer resolution with potential usefulness to studies of genetically altered mice.

The silicon surface of commercial atomic force microscopy (AFM) probes loses its hydrophilicity by adsorption of airborne and package-released hydrophobic organic contaminants. Cleaning of the probes by acid piranha solution or discharge plasma removes the contaminants and renders very hydrophilic probe surfaces. Time-of-flight secondary-ion mass spectroscopy and X-ray photoelectron spectroscopy investigations showed that the native silicon oxide films on the AFM probe surfaces are completely covered by organic contaminants for the as-received AFM probes, while the cleaning methods effectively remove much of the hydrocarbons and silicon oils to reveal the underlying oxidized silicon of the probes. Cleaning procedures drastically affect the results of adhesive force measurements in water and air. Thus, cleaning of silicon surfaces of the AFM probe and sample cancelled the adhesive force in deionized water. The significant adhesive force values observed before cleaning can be attributed to formation of a bridge of hydrophobic material at the AFM tip-sample contact in water. On the other hand, cleaning of the AFM tip and sample surfaces results in a significant increase of the adhesive force in air. The presence of water soluble contaminants at the tip-sample contact lowers the capillary pressure in the water bridge formed by capillary condensation at the AFM tip-sample contact, and this consequently lowers the adhesive force.

This paper reports on thermal strain analysis of integrated circuit (IC) packages using the optical, atomic force microscope (AFM), and scanning electron microscope (SEM) Moiré methods. The advantages and disadvantages of a full field optical Moiré, a micro‐optical Moiré, AFM Moiré, and SEM Moiré methods are compared. The full field Moiré interferometry is used to investigate the deformations and strains induced by thermal loading in various packages at the macrolevel. The micro Moiré interferometry is used to study the strains in the small solder joints. An optical Moiré interferometer with a mini thermal‐cycling chamber can be used for real time measurements of thermal deformations and strains of IC packages under thermal testing. Furthermore, the novel methods, AFM Moiré and SEM Moiré, can be also utilized to measure thermally induced deformations and strains of IC packages conveniently using the equipment that is commonly and primarily used for many other applications.

This thesis deals with the development of microfabrication technologies based on photoplastic structuring by lithographic and molding techniques. These technologies, combined with an original releasing method, allow for the simple fabrication of pseudo three-dimensional, soft, photoplastic microstructures with features and shapes that are difficult to obtain with standard micro-machining. The photoplastic material used is the SU-8 photoresist, a material that is used increasingly in the growing field of micro-electro-mechanical systems (MEMS). Photoplastic SU-8 microstructure fabrication is based on a combination of multi-layer spin-coating, molding and photolithographic processing of the resist on a prestructured silicon substrate. The final product is obtained by release of the structure from the substrate. This thesis describes several SU-8 microstructures which have been developed with an emphasis on atomic force microscopy (AFM) and scanning-near field optical microscopy (SNOM) probes. Scanning probe microscopy is a well-established technique for surface analysis, but batch-fabricated, low-cost probes still remain a challenging issue. Using a polymer for the cantilever facilitates the realization of mechanical properties that are difficult to achieve with classical silicon technology. The design, fabrication and testing of single lever and cassettes of multiple single-lever probes are presented and demonstrates the potential for fabrication structures with complicated shape and features. The fabrication process for SU-8 AFM probes is a simple batch process in which the integrated tips and the levers are defined in one photolithography step. Tip radii of curvature smaller than 15 nm have been obtained. The cantilever thickness depends on the spin-coating parameters, hence it can be very well controlled over a full wafer. Photoplastic cantilevers with thicknesses ranging from 1 to 6 μm have been produced. Imaging soft, condensed matter with photo-plastic levers, which uses laser beam deflection sensing, exhibits a resolution that compares well with that of commercially available silicon cantilevers. Lateral resolution of 5–6 nm has been estimated from imaging DNA-plasmid molecules. A vertical resolution of the order of 0.1 nm has been found. A similar fabrication technique was also developed to fabricate photo-plastic tips for SNOM that are to be attached to optical fibers. This technique allows optical apertures to be integrated at the end of the well-defined tip directly by probe fabrication, without the need for any post-processing for the aperture formation. Sub-100 nm aperture have been fabricated using this technique. Simple fabrication, as well as topographical and optical imaging demonstrate the potential of photoplastic-based probes for both AFM and SNOM applications, as well as for future combined probes development. In addition, the fabrication of functional microstructures by using SU-8 processing needs to be combined with other microfabrication techniques. Simple releasing of the molded structures from the substrate is especially of great importance. A sacrificial layer technique based on electrochemical etching enhancement has been developed and combined with the fabrication of the different photoplastic SU-8 probes presented. This technique allows the fast releasing of large microstructures and has been demonstrated by releasing other SU-8 photoplastic microfabricated devices.

Strong bond between the gold wire and the silver-plated leadframe is significantly crucial for maintaining either bondability or reliability during IC (integrated circuit) manufacturing process and IC application in the fields. This study investigated the surface and grain structure of the silver-plated film on the copper leadframe in terms of surface roughness test by atomic force microscopy (AFM), thickness, hardness measurement and grain structure by transmission electron microscopy (TEM). The characteristics of the silver-plated film surface and structure, e.g., surface roughness, hardness, film thickness and grain structure impacted the bonding quality which was verified by the full factorial design of experiment (DOE) wedge pull test. Implications of AFM, TEM and DOE results showed that more smooth and softer crystal surface and fewer grain boundaries of silver-plated film provided the higher pull strength which improved the yield of wedge bondability and reliability.

IC failure analysis: Techniques and tools for quality and reliability improvement

A scanning capacitance microscope (SCM) combines a differential capacitance measurement with an atomic force microscope (AFM). The AFM controls the position and contact force of a scanning probe tip, and a sensor simultaneously measures the capacitance between that tip and the sample under test. The capacitance sensor can detect very small (≈10 −21 F) changes in capacitance. The sensor is electrically connected to an AFM cantilevered tip that has been coated with metal to make it electrically conductive. As the SCM/AFM tip is scanned across the sample surface, simultaneous images of topography and differential capacitance are obtained. Currently, the most common application of SCM is imaging the carrier concentration variations (also known as dopant profiles) qualitatively within transistors that are part of silicon integrated circuits (ICs). Another application that has encouraged the development of commercial SCMs is the need for quantitative two‐dimensional (2‐D) carrier profiling of silicon transistors. This article reviews the history, operating principles, and applications of SCM. Special emphasis is placed on measuring quantitative 2‐D carrier profiles in silicon using SCM. This article also reviews implementation and applications of the scanning capacitance spectroscopy (SCS) technique, the intermittent contact mode of SCM (or IC‐SCM), and two other techniques that can also be classified as scanning capacitance probes: the scanning Kelvin probe microscope (SKPM), when operated at twice its drive frequency, and the scanning microwave microscope (SMWM).

the paper presents a failure analysis method of carbon/epoxy composite loops with the assistance of computer. The composite loops wrapped around metal pins form a kind of mechanically-fastened joint by filament winding technique. As are well known, mechanically-fastened joints under tensile loads generally fail in three basic modes, referred to as net-tension mode, shear out mode, and bearing mode. Among them, the bearing mode is an ideal failure mode. Through the failure analysis, the main failure mode of the joint of carbon/epoxy composite loops wrapped around metal pins mainly is the bearing mode. The failure load and the failure mode of the joint were analyzed numerically and experimentally. Hashin failure criteria were used in the failure analysis. Experiments of composite loops proved that the main failure mode of loops was bearing and delamination which were coincident with the numerical analysis.

This paper presents the design and implementation of a novel control architecture with a resonant controller, an integral controller, and a velocity feedback controller to improve the high speed imaging performance of a nanopositioning stage, piezoelectric tube scanner (PTS) used in most of the commercial atomic force microscopes (AFMs). The design of the controller is done by considering the lateral and longitudinal axes of the PTS as an independent single-input single-output (SISO) system. The controller proposed in this paper is able to achieve a bandwidth close to the first resonance frequency of the PTS and the controller is robust against the changes in resonance frequency of the scanner due to the load change on the scanner. The performance of the proposed controller is illustrated by comparing with an integral controller and it is shown that the bandwidth increased by the proposed controller is five times greater than the bandwidth that can be achieved by using the integral controller in a commercial scanner. Experimental images with the open-loop, closed-loop and built-in AFM proportional integral controller are presented at scanning rates of 15.62 Hz, 31.25 Hz, 62.5 Hz, and 125 Hz to demonstrate the advantage of the proposed controller.

The atomic force microscope (AFM) has previously been applied to the measurement of surface forces (including adhesion and friction) and to the investigation of material properties, such as hardness. Here we describe the modification of a commercial AFM that enables the "stiffness" of interaction between surfaces to be measured concurrently with the surface forces. The stiffness is described by the rheological phase difference between the response of the AFM tip to a driving oscillation of the substrate. We present the interaction between silica surfaces bearing adsorbed polymer, however, the principles could be applied to a wide variety of materials including biological samples.

We report parallel topographic imaging and nanolithography using heated microcantilever arrays integrated into a commercial atomic force microscope (AFM). The array has five AFM cantilevers, each of which has an internal resistive heater. The temperatures of the cantilever heaters can be monitored and controlled independently and in parallel. We perform parallel AFM imaging of a region of size 550 μm × 90 μm, where the cantilever heat flow signals provide a measure of the nanometer-scale substrate topography. At a cantilever scan speed of 1134 μm s−1, we acquire a 3.1 million-pixel image in 62 s with noise-limited vertical resolution of 0.6 nm and pixels of size 351 nm × 45 nm. At a scan speed of 4030 μm s−1 we acquire a 26.4 million pixel image in 124 s with vertical resolution of 5.4 nm and pixels of size 44 nm × 43 nm. Finally, we demonstrate parallel nanolithography with the cantilever array, including iterations of measure-write-measure nanofabrication, with each cantilever operating independently.

Although photon emission microscope (PEM) systems are widely used in integrated circuit failure analysis, there is no known quantitative baseline to assess and compare the overall sensitivity performance of PEM systems. This paper describes a method to quantify the overall sensitivity of PEM systems based on spectral detectivity measurements. It has been applied to HgCdTe (MCT) and InGaAs PEM systems. It is also applied to an InGaAs PEM system to quantify the change in the detectivity of the InGaAs PEM system as the temperature of the detector changes. The method is also used to compare the signal to noise ratio of an emission image by normal time integration with digital integration where many frames of an emission image is added up to produce a single emission image